Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
The International Education Journal: Comparative Perspectives Vol. 16, No. 4, 2017, pp. 29-46 https://openjournals.library.sydney.edu.au/index.php/IEJ
29
What contributes more to the ranking of higher education institutions? A comparison of three
world university rankings
Ya-Wen Hou
National Taichung University of Education, Taiwan:
W James Jacob
University of Memphis, US: [email protected]
Recently, many universities have drawn attention to world university
rankings, which reflect the international competition of universities and
represent their relative statuses. This study does not radically contradict
types of global university rankings but calls for an examination of the effects
of their indicators on the final ranking of universities. By using regression
analysis, this study investigates the indicator contribution to the ranking of
universities in world university ranking systems, including the Academic
Ranking of World Universities (ARWU), Times Higher Education (THE),
and QS World University Rankings. Results show that in the ARWU system,
three indicators regarding faculty members who won Nobel Prizes and
Fields Medals and papers published in Nature and Science and in the
Science Citation Index and Social Science Citation Index journals predicted
the ranking of universities. For the QS and THE systems, the more powerful
contributors to the ranking of universities were expert-based reputation
indicators.
Keywords: world university rankings; ranking indicators; indicator
contribution; ranking of universities; university position
INTRODUCTION
Driven by globalization and massification in higher education (Altbach, 2012),
university rankings and league tables are having an ever greater impact on higher
education institutions (HEIs). Similar to the pursuit of accountability and objective
evaluation, university rankings exist ubiquitously (Wildavsky, 2010). Both national and
international university rankings are growing explosively and becoming more
specialized by, for example, focusing on research performance or institutional reputation
(see Rauhvargers, 2011; Shin & Toutkoushian, 2011). In particular, world university
rankings, which are our concern in this paper, are considered by many to be a means of
representing academic excellence and increasing prominence of HEIs in both local and
global contexts.
Hou & Jacob
30
Improving global rankings 1 in league tables is often a priority goal for many
universities. World university rankings serve as a reference point for student choices for
universities and scholar mobility across the globe, provide a guide to public policies,
help in decision-making by funding agencies and university leaders, and even play a role
in positioning and measuring the performance of higher education institutions in the
domestic and global contexts (Altbach, 2006, 2012; Bastedo & Bowman, 2011;
Hazelkorn, 2009, 2014; Huisman & Currie, 2004; Marginson & van der Wende, 2007;
Salmi & Saroyan, 2007; Williams, 2008). In light of the positive relationship between
Web links and ranking orders found by Lee and Park (2012), universities themselves
endeavour to participate in global ranking activities and pursue higher ranks to obtain
greater visibility and resources from multiple stakeholders (Hazelkorn, 2014). Thus,
global rankings are often regarded as “a mechanism of agenda setting” with soft power
(Lo, 2011, p. 216) and “an integral part of [the] status culture” of higher education
competition (Marginson, 2014, p. 45). The higher the ranking, the more visibility and
opportunities HEIs generally gain within their respective countries and across the globe.
World university rankings are an attractive and often competitive measurement of
institutional performance by bibliometric methods (van Raan, 2005). To some extent,
global rankings value stakeholder choices and investments, set institutional benchmarks,
reorganize higher education institutions that work ineffectively, determine institutional
priorities, and boost faculty academic professional reputation (Hazelkorn, 2009; Shin &
Toutkoushian, 2011). An empirical study conducted by Bastedo and Bowman (2011)
links college rankings with an institutional ability to gain greater financial resources. In
terms of student recruitment, global rankings play an important role in student
preferences and choice. A report initiated by the QS Intelligence Unit (2015) notes that
over 70% of surveyed students consider global rankings more important in their
university selection process than national or regional rankings, making them a crucial
factor in the institutional selection for many students (Roberts & Thomson, 2007)
because students tend to relate higher ranked institutions with better reputation and
academic excellence.
World university rankings also influence strategic direction and decisions made by
senior higher education administrators, including in how they react among and between
leaders of other HEIs (Hazelkorn, 2009). Higher ranked universities are like institutional
sponges that generally have greater opportunities to gain sustained public funding and
private investments. Institutional reputation linked to global rankings also makes it
easier for the top-ranked HEIs to attract scholars and students from domestic and
international locations. World university rankings serve as an important underpinning of
institutional reputation (Bowman & Bastedo, 2011) and in providing greater perceived
“credibility” (Vieira & Lima, 2015, p.63). Rankings are also influential in producing a
non-negligible effect on graduates’ wages (Carroll, 2014). Many HEIs strive to align
strategic plans and institutional performance to the criteria of world university rankings
to solidify and boost their ranking position among the top institutions.
However, world university rankings have raised controversy, including their neglect of
audiences’ needs, the preference for English-language publications as a key indicator, an
overemphasis on the fields of science and medicine, subjectivity of survey indicators,
arbitrary weighting, the variability of ranking results, and the bias between
1 In this article, the term global rankings refers to world university rankings produced on an annual basis by several leading ranking systems.
What contributes more to the ranking of higher education institutions?
31
ordinal/numeric representation and the actual quality of university education (Bastedo &
Bowman, 2011; Dill & Soo, 2005; Fidler & Parsons, 2008; Frey & Rost, 2010;
Marginson, 2014; Proulx, 2007; Saisana, d’Hombres, & Saltelli, 2011; Taylor &
Braddock, 2007; Tofallis, 2012; van Vught & Westerheijden, 2010; Williams, 2008).
Furthermore, an overemphasis on world university rankings is like jumping into a risky
venture; for instance, rather than focusing their decision on which institution to attend
based on outstanding academic performance, students often make their choice on
institutional reputation (Taylor & Braddock, 2007).
Another common indicator critique of world university rankings is the preference for
research publications from the Science Citation Index (SCI) and the Social Science
Citation Index (SSCI). SCI and SSCI are widely recognized in academic circles for
defining success, boosting institutional reputation, and justifying university rankings.
But overvaluing the SCI and SSCI indicator may give rise to what Su (2014, p. 51) calls
an “I-idolization” or an overemphasis on the leading publication indices. Other scholars
list several shortcomings that relate to academic recognition, the marginalization of the
humanities and social sciences, and institutional image (Deem, Mok, & Lucas, 2008;
Delgado & Weidman, 2012).
There is also an invisible pressure in the pursuit of increasing institutional reputation
that often intensifies the competition between HEIs and between countries. Moreover,
Proulx (2007) argued that ranking results based primarily on SCI and SSCI research
outputs are likely to persuade many leaders of ranked universities to over-emphasize the
need for greater research publication outputs rather than focusing on developing relevant
strategies to become world-class universities. Thus, world university rankings often lead
to in an inherent risk of competition that ultimately excludes many of flagship
universities (Amsler & Bolsmann, 2012; Douglas, 2016); HEIs compete internationally
for human and financial resources, and their competitive institutional behaviours are
simultaneously reinforced by the global ranking results (van Vught & Westerheijden,
2010).
Although many studies have documented various issues surrounding global university
rankings, few studies have demonstrated the relationship between the indicators used
and the ranking of universities in a particular ranking system; that is, which indicators
have a greater impact on determining the ranking of universities. Understanding the
contribution of indicators of global rankings is fundamental to understanding the role of
global rankings and their methodologies as well as HEIs’ strategies for pursuing global
rankings.
This paper reports on a study investigating indicator contributions to the ranking of
institutions of three of the most prominent world university ranking systems: the
Academic Ranking of World Universities (hereafter referred to as ARWU) developed
by Shanghai Jiao Tong University in China; the Times Higher Education World
University Rankings (hereafter referred to as THE) created by Thomson Reuters; and
Quacquarelli Symonds’ World University Ranking (hereafter referred to as QS). In other
words, our study sought to explore whether the weights of the indicators in these three
global ranking systems are different to the assigned weights shown in their
methodologies, and whether some indicators matter more than others. In this study, we
first describe the characteristics and methodologies of the three ranking systems and
common criticisms regarding their indicators. We then analyse the indicators’
contribution to the ranking of HEIs to better understand the implications global
university rankings have in practice.
Hou & Jacob
32
THREE WORLD UNIVERSITY RANKING SYSTEMS:
FEATURES AND CRITICISM
The ARWU, THE, and QS rankings are the “big three,” according to Hazelkorn (2014,
p. 17), being among the most frequently used by scholars, administrators, policy makers,
and students. The first global ranking system developed was the ARWU in 2003. The
next year, Times Higher Education and the Quacquarelli Symonds Company co-
published their own ranking systems, which is usually referred to as THE-QS (Liu &
Cheng, 2011). However, in 2010, THE, and QS ended their collaboration and separated
into two separate ranking systems.
Features of the selected ranking systems
Table 1 shows the background of these three systems.
Table 1: Main characteristics of the three university ranking systems
Academic Ranking of
World Universities
(ARWU)
QS World University
Ranking
Times Higher
Education World
University Ranking
(THE)
Background
Issued by
Academic institution
(Shanghai Jiao Tong University)
Media
(Quacquarelli Symonds)
Media
(Thomson Reuters)
Years 11 (since 2003) Since 2004, THE cooperated with QS. However, THE decided to change the partner and developed its own methodology in 2010
Target audience No No No
Methodology
Criteria/Dimensions 4 0 5
Number of indicators 6 6 13
Conducting reputation survey
No Yes Yes
Data sources
Thomson Reuters' Web of Science Database, Resources of National agencies
Scopus Database, University portfolio, Survey
Thomson Reuters' Web of Science Database, University portfolio, Survey
Results published on the web
Yes Yes Yes
Ordinal Results
Top 500
(Single ranks to 100 and then groups)
Top 700
(Single ranks to 400 and then groups as 401–410, 411–420, 421–430, 431–440, 441–450, 451–460, 461–470, 471–480, 481–490, 491–500, 501–550, 551–600, 601–650, 651–700)
Top 400
(Single ranks to 200 and then groups as 201–225, 226–250, 251–275, 276–300, 301–350, 351–400)
Source: Authors.
ARWU is created by an academic institution (Shanghai Jiao Tong University), while the
other two are developed by the mass media. All the selected ranking systems focus on
the evaluation of research-led universities worldwide, even though their methodologies
are not similar.
What contributes more to the ranking of higher education institutions?
33
Rather than the specific target groups, all individuals and stakeholders engaged in
higher education are the intended audience. These global rankings are likely to influence
and drive the perceptions and behaviours of individuals and organizations, such as
students, parents, faculty, and staff members, public authorities, and employers and
community members (Thakur, 2007; van Vught & Westerheijden, 2010).
Except for surveys and resources from national agencies and university profiles, all the
selected ranking systems use databases to analyse research publications and citations
through a bibliometric method. QS uses the Scopus database, and the other two collect
information from Thomson Reuters’ Web of Science database. The ARWU system also
collects data from select websites (e.g., SCI, SSCI, Nobel laureates, and Fields Medals),
and THE and QS also conduct reputation surveys.
The three ranking systems commonly publish their results online using ordinal rankings
in lists. ARWU publishes a list with the top ranked 500 institutions in which
universities are ranked from one to 100 and then grouped as 101–150, 151–200, 201–
300, 301–400, and 401–500. The QS system uses the methodological framework that
served as the original version of the THE-QS rankings (Quacquarelli Symonds, 2014)
and publishes a list of the top 700 universities online, of which the top 400 institutions
are singly ranked and the latter 300 are grouped. THE releases an online league table of
the top 400 universities, which are singly ranked up to 200 and then grouped as 201–
225, 226–250, 251–275, 276–300, 301–350, and 351–400. The ARWU system has
better ordinal proportionality than the QS and THE systems (Marginson, 2014) because
of its fixed proportion of ordinal results.
Evaluative standards of the selected ranking systems
Each system uses its own standards for evaluation and weighting. In 2011, THE ended
its collaboration with QS and developed a new methodology with a different partner—
Thomson Reuters. Instead of the old six indicators that QS still uses, the new THE
methodology consists of 13 indicators ranging from teaching and research to knowledge
transfer (Thomson Reuters, 2010). Table 2 illustrates the indicators and their assigned
weights in these ranking systems.
The ARWU ranking system includes six indicators among four dimensions (Shanghai
Jiao Tong University, 2014). First, the dimension of education quality is determined by
one indicator—the number of alumni who have won Nobel Prizes and Fields Medals
(coded as Alumni), and this indicator contributes 10% to the overall score. Second, the
dimension of faculty quality is evaluated by two indicators; one is related to Nobel
Prizes and Fields Medals granted to faculty members (coded as Award), and the other is
HiCi, a parameter related to highly cited researchers in 21 subject categories. These two
indicators account for 20% each. Third, the research output dimension is also
determined through two indicators: papers published in Nature and Science (coded as
NS) and those indexed in SCI and SSCI (coded as PUB). These two indicators account
for 20% each. Finally, the per capita performance of an institution (abbreviated to PCP)
contributes 10% to the overall score.
Hou & Jacob
34
Table 2: Indicators and assigned weights of selected university ranking systems
Title ARWU QS THE D
imen
sio
n/
Ind
ica
tor
Quality of Education
- Alumni (10%)
Quality of Faculty
- Award (20%)
- HiCi (20%)
Research Output
- Nature and Science (20%)
- PUB: SCI & SSCI (20%)
Per Capita Performance (10%)
Academic reputation (40%)
Employer survey (10%)
Citation per faculty (20%)
Faculty-student ratio (20%)
International students (5%)
International faculty (5%)
Teaching (30%)
- Reputation survey for teaching (15%)
- Staff-student ratio (4.5%)
- Doctoral-bachelor's ratio (2.25%)
- PhDs awarded (6%)
- Institutional income per faculty member (2.25%)
Research (30%)
- Reputation survey for research (18%)
- Research grants (6%)
- Papers in peer-reviewed journals (6%)
Citation impact (30%)
Industry income (2.5%)
International outlook (7.5%)
- Ratio of international-domestic students (2.5%)
- Ratio of international-domestic staff (2.5%)
- Publication with international co-authors (2.5%)
Source: Created by the authors with criteria from Quacquarelli Symonds (2014), Shanghai Jiao Tong University (2014), and Thomson Reuters (2014).
The THE system uses 13 indicators for five dimensions (Thomson Reuters, 2014). First,
the teaching dimension is assigned a weight of 30% and is determined through five
indicators, teaching reputation survey, staff-to-student ratio, doctorate-to-bachelor ratio,
doctorate awards by an institution, and institutional income scaled against academic
staff numbers. Second, the research dimension has a 30% share and is established
through research reputation survey, research grants, and the number of papers published
in academic journals. The third dimension is citation impact, given a weight of 30%.
The fourth dimension is research funding from industry, contributing 2.5% to the overall
score. Finally, the international outlook dimension of an institution is assigned a weight
of 7.5% and is determined through the international-to-domestic student ratio,
international-to-domestic staff ratio, and number of internationally co-authored research
papers.
The QS ranking system still uses the original methodological framework (Quacquarelli
Symonds, 2014). Among the six indicators included in the QS system, the most
important is the academic peer reputation survey, with a weight of 40%. Another
reputation survey addresses employers and contributes 10% to the ranking. Then, the
two indicators of citations per faculty and faculty-student ratio contribute 20% each to
What contributes more to the ranking of higher education institutions?
35
the overall score. Finally, the numbers of international students and faculty indicators
have a weight of 5% each.
It is important to note the similarities in these ranking systems; the differences lie in the
weights of the indicators used. University rankings, according to Proulx (2007), should
represent the characteristics of the ranked universities and “avoid a one-size-fits-all
typology” (p. 76). All the indicators of these ranking systems are grouped into five
categories, including teaching, research, service mission, reputation management, and
internationalization of higher education institutions (Table 3). These five categories
result from several multifaceted and interactive factors. In particular, the last two
categories seem to be relevant in the context of knowledge-based economic societies.
However, the weights assigned to the indicators seem to reflect the emphasis of each
global ranking system and have some biases. For instance, the ARWU system focuses
on research and eliminates teaching, service mission, and internationalization; and the
indicators of the QS and THE systems are incomplete in assessing the effectiveness of
teaching and learning and research funding (Marginson, 2014), though both have some
measures for these five categories.
As shown in Table 3, the ARWU system emphasizes research excellence, while the QS
system focuses more on universities’ reputation (Aguillo, Bar-Ilan, Levene, & Ortega,
2010; Huang, 2011). For instance, the ARWU system assigns a weighting of
approximately 90% to impressive research performance, such as Nobel Prizes and
Fields Medals granted to alumni and faculty members and publication in famous
English-language journals. By contrast, the QS system depends greatly on university
reputation, representing nearly 50% of the total score.
Table 3: Priorities of selected university ranking systems
Title
Category
ARWU QS THE
Teaching
Research (*) (*)
Service
Reputation (*)
Internalization
Note: * refers to the category given the most assigned weights in the system. Source: Authors.
In addition, the dimensions evaluated by the QS and THE ranking systems are similar,
but the indicators of the THE system are more detailed and complex (Marginson, 2014).
In the THE system, more than one-third of the overall score is associated with research
outcomes such as research grants, publications, and citations. However, in the QS
system, less than one-fourth of the total weight is allocated to research outcomes. Even
though both of these two ranking systems measure university reputation, the THE
system assigns approximately 33% to reputation surveys, while, in the QS system, it
accounts for 50% of the overall score. Both ranking systems give approximately 10% to
the internationalization of higher education institutions and have some indicators to
assess the teaching mission, but the QS system employs only one indicator of it, the
faculty/student ratio.
Hou & Jacob
36
Criteria debate of the selected ranking systems
As already noted, the three ranking systems have been extensively criticized (see
Marginson, 2014; Taylor & Braddock, 2007). The incomplete databases used by the
three ranking systems tend to have biases against the non-English-language publications
and the fields of social science and humanities (Amsler & Bolsmann, 2012; van Raan,
2005). The biases in favour of hard-science and English-language publications also
result from the different publication cultures and citation habits in diverse fields (Frey &
Rost, 2010; Saisana et al., 2011; van Vught & Westerheijden, 2010).
Another similar criticism of the three ranking systems is the challenge concerning their
selections of indicators and arbitrary weighting. The numeric, comparable, and
standardized league tables produced by global university rankings often lead the
uninformed public to believe the truth of the information. Through league tables,
everyone can easily interpret and compare the quality of certain universities. Those who
publish rankings also believe that the ranking results reflect the position and quality of a
university through a rigorous and objective process of evaluation (Rauhvargers, 2011).
However, as argued by Williams and Van Dyke (2008), “the objectivity does not ensure
that the measures actually chosen are always appropriate” (p. 2). Taylor and Braddock
(2007) argued that the weights of indicators depend on the significance of the set of
indicators suggested by higher education consultants, and most ranking systems have
chosen the “suitable” indicators instead of the negative ones, jeopardizing institutional
or national interest (Marginson, 2014, p. 46). For instance, because of the initial purpose
of understanding the global standing of top Chinese universities (Liu & Cheng, 2011),
the ARWU system relies heavily on research performance without consideration for the
teaching, social service, internationalization, and employability of university graduates
(Marginson, 2014; Saisana et al., 2011).
In the ARWU ranking system, focusing on research-oriented indicators is frequently
criticized. Aside from the biases in favour of hard-science and English-language
journals, the Nobel indicator affects the lower-ranked universities located in countries
with few Nobel Prize and Fields Medal winners, and underrepresents the diversity of
academic fields and other scholars’ achievements (Huang, 2011; Marginson, 2014).
For the QS and THE systems, the major criticism involves the subjectivity of reputation
surveys, the teaching indicators, and the instability of the rankings. Employing expert-
based reputation surveys as a ranking indicator is subjective to the bias caused by
human opinions and judgments on a university (Salmi & Saroyan, 2007; Williams &
Van Dyke, 2008). In other words, the subjectivity of survey indicators is inevitable in
relatively objective ranking indicators (Dill & Soo, 2005; Taylor & Braddock, 2007). In
terms of teaching criteria, Trigwell (2011) stated that, in the QS system, using a single
indicator—staff-to-student ratio—to assess the teaching performance of a university is
questionable; this indicator depends on class size but also cannot accurately reflect
teaching quality and the diversity of teaching and learning activities. As with the QS
system, it is difficult to evaluate actual teaching effectiveness even though the THE
system adds other indicators to assess teaching performance, such as PhDs, the doctoral-
bachelor's ratio, and the facilities and income of an institution. Then, the variability and
fluctuation of the rankings result from the frequency of changes in methodology and the
use of surveys in the QS and THE systems (Marginson, 2014; Saisana et al., 2011). The
empirical study conducted by Aguillo et al. (2010) indicated that the dissimilarities
between the THE-QS rankings for different years are high. In other words, the THE and
QS systems are more unstable than the ARWU system.
What contributes more to the ranking of higher education institutions?
37
Intuitively, all ranking systems have different impacts on ranking results, including the
overall score and the ordinal ranking of universities because of their different
frameworks of indicators. Investigating which indicators of the three world university
rankings best predicts the ranking of universities would be interesting.
RESEARCH METHOD
Data sources
All data were obtained from the ARWU, THE, and QS world university rankings. As
the source of data, we chose the top 100 universities from each selected ranking system
in accordance with their 2013-14 world university rankings released on their websites.
For each ranking system, the data we collected included the scores for every criterion
and the overall scores, as well as the ranking of the 100 institutions. However, because
Harvard University is the benchmark in the ARWU system, it was excluded from our
data from the ARWU rankings.
Data analysis
We used secondary data and regression analysis in the study. We used regression
analysis to explore the effects of independent variables on an outcome variable
(Treiman, 2009). For each ranking system, we used bivariate regression to examine the
effect of a single indicator on the ranking separately, and we employed multiple
regression to investigate the impact resulting from the whole set of the indicators.
Limitations and contributions
The major limitation of this study is the change of ranked universities that are shown on
the lists of the three world university rankings. In this study, we chose world university
rankings in 2013-14 as our data set. We chose to study only the top 100 ranked
institutions but those in the top 100 changed depending on the year, thus leading to a
bias. However, the variation of ranked universities on the top 100 lists of these three
global rankings is smaller than those on other ordinal categories. In order to eliminate
biases resulted from the uncertainty and variation, we focused on the top 100 in the
selected world university rankings.
However, two features of our study should be emphasized. First, although every ranked
university receives an overall score and a respective ranking, this ordinal ranking is
likely to represent the position of a university. Thus, in our study, we paid more
attention to the ranking of universities than to the final scores these HEIs received.
Second, we only selected the top 100 universities instead of all ranked schools shown in
the rankings (e.g., the top 500 in the ARWU rankings) because these top 100
universities are given unique rankings. Moreover, receiving the first-tier rankings
implies that these universities have more opportunities and better competitiveness than
others in terms of marketing.
RESULTS AND COMPARISON
Overall, all Pearson correlation coefficients between the single indicator and the ranking
of a university in each system were negative. The reason is that the increase of the
numerical value refers to more attention on indicators but not on the higher ranking of
Hou & Jacob
38
institutions. The smaller the value (such as Top1means) relates to institutions with the
best performance. Rather than indicating the relative significance between positive and
negative correlations, this shows that the indicators and the ranking of a university in
each system move in the opposite way (Treiman, 2009). According to the results of our
study, all correlation coefficients were statistically significant with respect to ranking of
universities in the ARWU and QS ranking systems, while some were nil in the THE
system. The following section describes the regression analysis of each selected ranking
system in detail.
ARWU system
Table 4 shows the regression summary of the ARWU rankings. The final model (model
7) that includes the six indicators explained 83% of the variance in the ranking of
universities (F (6, 92) = 73.403, p < .001). Even the adjusted R2 also provided an
explanation of 82%. As shown in Table 4, three indicators were significantly and
inversely related to the ranking of universities, including Award (b = -.542, p < .001),
NS (b = -.770, p < .001), and PUB (b = -.728, p < .001). By comparing their
standardized weights (βeta-weights), we found that the Award indicator had the most
substantial impact on the ranking of universities (β = -.405), more than the NS (β = -
.362) and PUB (β = -.294) indicators.
Table 4: Regression analysis for the ARWU system
Model (with indicators) R R2 Adjusted R2
1 Alumni .615 .378 .372
2 Award .697 .486 .481
3 HiCi .766 .587 .583
4 NS .838 .702 .699
5 PUB .590 .348 .341
6 PCP .453 .205 .197
7 Alumni, Award, HiCi, NS, PUB, PCP .910 .827 .816
b-weight (βeta-weight)
Indicator Between each indicator and the rank Final model
Alumni -1.066*** (-.615)
-.066 (-.038)
Award -.932*** (-.697)
-.542 (-.405)***
HiCi -1.646*** (-.766)
-.200 (-.093)
NS -1.783*** (-.838)
-.770 (-.362)***
PUB -1.462*** (-.590)
-.728 (-.294)***
PCP -1.091*** (-.453)
.141 (.058)
Notes: (a) ***p ≤ .001, **p ≤ .01, *p ≤ .05. (b) Alumni = Alumni winning Nobel Prizes and Fields Medals; Award = Staff winning Nobel Prizes and Fields Medals; HiCi = Highly cited researchers; NS =Articles published in Nature and Science; PUB = Science Citation Index and Social Science Citation Index; PCP = Per capita academic performance.
What contributes more to the ranking of higher education institutions?
39
Source: Authors.
Using bivariate regression, each indicator had statistical significance in explaining its
effect on the ranking of universities. The top two contributors to institutional ranking
were the NS and HiCi indicators (models 4 and 3), which could individually explain
more than 50% of the variance in the ranking of universities, but unfortunately, the HiCi
indicator was not statistically significant in the final model (model 7). The third most
influential contributor to institutional ranking was the Award indicator, with an adjusted
R2 of 48%. In other words, 48% of the variation in the ranking of universities could be
explained with Award. Then, the Alumni and PUB indicators separately explained
approximately 35% of the variance in the ranking system, but the PUB indicator had
statistical significance in the final model. Finally, the PCP indicator received an
adjusted R2 value lower than 20%; relatively, the PCP indicator contributed less to the
ranking of universities.
THE ranking system
The overall model (model 6) that includes the five criteria explained 90% of the
variance in the ranking of universities (F (5, 85) = 164.782, p < .001). As the results in
Table 5 illustrate, three indicators—teaching (b = -.697, p < .001), research (b = -.775, p
< .001), and citation (b = -.490, p < .001)—were significantly and inversely related to
the ranking of universities and international outlook (b = -.124, p < .05), while the
industry income criteria had no significance. In comparing their standardized weights
(βeta-weights), we found that the research indicator had the most powerfully substantial
impact on the ranking of universities (β = -.506), and the second one was teaching (β = -
.400). Both of them were more than twice that of the citation indicator (β = -.206) and
more than four times that of the international outlook indicator (β = -.084).
Table 5: Regression analysis for the THE system
Model (with indicators) R R2 Adjusted R2
1 Teaching .904 .817 .815
2 Research .902 .813 .811
3 Citation .410 .168 .160
4 Int’l Outlook .127 .016 .006
5 Industry .113 .013 .002
6 Teaching, Research, Citation, Int’l Outlook, Industry .952 .906 .901
b-weight (βeta-weight)
Indicator Between each indicator and the rank Final model
Teaching -1.586*** (-.904)
-.697 (-.400)***
Research -1.398*** (-.902)
-.775 (-.506)***
Citation -.991*** (-.410)
-.490 (-.206)***
Int’l Outlook -.193 (-.127)
-.124 (-.084)*
Industry -.142 (-.113)
.014 (.011)
Notes: (a) ***p ≤ .001, **p ≤ .01, *p ≤ .05. (b) Int’l Outlook = International outlook; Industry = Industry income.
Source: Authors.
Hou & Jacob
40
When analysing the relationship between single indicator and the ranking of
universities, we found that three of five criteria had statistical significance in explaining
their effect on the ranking except the international outlook and industry income criteria.
The top two contributors to the ranking were teaching and research (models 1 and 2),
which could individually explain approximately 80% of the variance in the ranking of
universities. The third contributor to the ranking was the citation indicator, with an
adjusted R2 of 16%. The other two indicators had adjusted R2 values lower than 1%.
Interestingly, the international outlook indicator was statistically significant in the final
model but not important when we assessed its single effect on the ranking.
QS Ranking System
The regression results for the QS ranking system are illustrated in Table 6.
Table 6: Regression analysis for the QS system
Model (with indicators) R R2 Adjusted R2
1 Peer .746 .556 .552
2 Employer .535 .286 .279
3 F/S ratio .630 .396 .390
4 Int’l faculty .205 .042 .032
5 Int’l student .327 .107 .098
6 Citation .336 .113 .104
7 Peer, Employer, F/S ratio, Int’l faculty, Int’l student, Citation
.985 .970 .968
b-weight (βeta-weight)
Indicator Between each indicator and the rank Final model
Peer -2.035*** (-.746)
-1.456 (-.533)***
Employer -.974*** (-.535)
-.255 (-.140)***
F/S ratio -.751*** (-.630)
-.596 (-.500)***
Int’l faculty -.192* (-.205)
-.101 (-.108)***
Int’l student -.355*** (-.327)
-.161 (-.148)***
Citation -.506*** (-.336)
-.499 (-.332)***
Notes: (a) ***p ≤ .001, **p ≤ .01, *p ≤ .05. (b) Peer = Academic reputation by peer review; Employer = Employer survey; F/S ratio = Faculty/Student ratio; Int’l faculty = Proportion of international faculty; Int’l student = Proportion of international student.
Source: Authors.
The final model (model 7) that includes the six indicators provided a very strong
explanation of 97% for determining the ranking of universities (F (6, 93) = 497.673, p <
.001). As shown in Table 6, all indicators were significantly and inversely related to the
ranking of universities. By comparing their standardized weights (βeta-weights), we
found that the top two indicators with the most substantial impact on the ranking of
What contributes more to the ranking of higher education institutions?
41
universities were peer review (β = -.533) and faculty-student ratio (β = -.500). The third
most influential contributor was the citation indicator (β = -.332). The influence of these
three indicators was more than twice that of the other three indicators, including
employer survey and the number of international students and faculty members.
According to the bivariate regression, each indicator had statistical significance in
explaining its effect on the ranking. The most influential contributor to the ranking was
peer review (model 1), which could explain 55% of the variance in the ranking of
universities. The next most influential contributor to the ranking was the faculty-to-
student ratio, which could explain 39% of the variation in the ranking of universities.
The third one was the employer survey indicator, with an adjusted R2 of 28%. In other
words, 28% of the variation in the ranking of universities could be explained by the
employer survey. The other three indicators had adjusted R2 values equal to or less than
10%.
Most indicators of these three ranking systems made substantial contributions to the
ranking of the top universities except two indicators of the THE system: international
outlook and industry income. In the ARWU system, three indicators: Award, NS, and
PUB, had statistical significance in predicting the ranking of universities. Even though
the variance in the ranking of universities could be explained with HiCi (adjusted R2 of
58%), it was not statistically significant. That is to say, the ARWU ordinal ranking
could be determined by Award, NS, and PUB. In terms of the QS and THE systems, the
more powerful contributors to the ranking of universities were expert-based reputation
indicators, including the peer review of the QS system and the teaching and research
criteria of the THE system. These findings were consistent with the results of several
previous studies (e.g., Huang, 2011; Marginson, 2014; Saisana et al., 2011). This
seemed to imply that universities have opportunities to receive higher rankings if they
have tangible, popular, and customer-appreciated research products and an excellent
reputation.
The regression analysis of these three ranking systems suggest that not all indicators in
each ranking system contribute equally to the prediction of their final ranking of
universities. In other words, several indicators could explain their respective rankings,
while some indicators might not make an authentic contribution of the ranking
prediction. Moreover, the importance of the most influential indicators to the final
ranking of universities could be dissimilar to their assigned weights in the
methodologies. For instance, according to the ARWU methodology, the NS indicator is
assigned a weight of 20%, but this single indicator can explain approximately 70% of
the variance in the ARWU ranking of universities. Although the final ranking of
universities results from multiple factors and are influenced by them, the effect of a
single indicator on the final ranking results cannot be neglected.
CONCLUSION
Facing increasing competition between HEIs in domestic and global contexts, the
number of ranking systems at the national and international levels is increasing.
University rankings are seen as a meaningful representation of bettering academic
excellence and institutional reputation. In order to achieve these goals, most HEIs make
a concerted effort to participate in institutional ranking activities rather than escape from
them. In particular, world university rankings have gradually drawn greater attention in
international and comparative higher education. The basic goal of global university
Hou & Jacob
42
rankings is to provide information to inform student choices of universities, but the
impact and use of global rankings have changed. Global university rankings serve as
tools for evaluating universities’ outstanding performance as well as marketing and
positioning within countries and around the world. They become politically
exclusionary instruments (Amsler & Bolsmann, 2012) and a tool of status control
(Marginson, 2014). In other words, the ordinal numbers shown in the global rankings
implies the position and competitiveness of a university. Thus, in the current study, our
intention is not to deny global university rankings but to argue that international ranking
systems should be carefully examined and that their results should be deliberately
interpreted.
The selected global ranking systems are not perfect in measuring higher education
institutional performance and in awarding their ordinal statuses across the globe. After
analysing the contributions of the indicators to the final ranking of universities in the
three ranking systems, we obtained several findings. First, most indicators of the three
ranking systems were positively correlated with their overall ranking except the
international outlook and industry income indicators of the THE ranking system. The
reason that these indicators were not statistically significant might involve their lower
assigned weights and whether universities were willing to provide accurate information
on financing and internationalization. Thus, we highlight the need for HEIs and those
who publish the rankings to be aware of and sensitive to the methodological issues and
the transparency of institutional financial and internationalization data.
We also note that, in the three ranking systems, not all indicators contribute to the
ranking of universities. This seems to imply that the ranking of HEIs might be
determined by a few indicators. The various methodologies of different ranking systems
may cause vulnerabilities in the seemingly objective evaluations and redundant
evaluative criteria. However, as stated by Rauhvargers (2011), readers seldom receive
and understand the actual information regarding the calculation to obtain the final
ranking. Unfortunately, too often readers might be misinformed. The audience might
also overestimate or underestimate the contributions of some indicators to the final
ranking. Hence, we caution higher education stakeholders at all levels against using and
interpreting the surface ordinal numbers and about making public decisions based solely
on global ranking systems. We also suggest that the indicators chosen for each ranking
system should be regularly examined to avoid redundant biases.
REFERENCES
Aguillo, I. F., Bar-Ilan, J., Levene, M., & Ortega, J. L. (2010). Comparing university
rankings. Scientometrics, 85(1), 243–256. doi: 10.1007/s11192-010-0190-z
Altbach, P. G. (2006). The dilemmas of ranking. International Higher Education, 42, 2–
3.
Altbach, P. G. (2012). The globalization of college and university rankings. Change:
The Magazine of Higher Learning, 44(1), 26–31. doi:
10.1080/00091383.2012.636001
What contributes more to the ranking of higher education institutions?
43
Amsler, S. S., & Bolsmann, C. (2012). University ranking as social exclusion. British
Journal of Sociology of Education, 33(2), 283–301. doi:
10.1080/01425692.2012.649835
Bastedo, M. N., & Bowman, N. A. (2011). College rankings as an interorganizational
dependency: Establishing the foundation for strategic and institutional accounts.
Research in Higher Education, 52(1), 3–23. doi:10.1007/s11162-010-9185-0
Bowman, N. A., & Bastedo, M. N. (2011). Anchoring effects in world university
rankings: Exploring biases in reputation scores. Higher Education, 61(4), 431–444.
doi: 10.1007/s10734-010-9339-1
Carroll, D. (2014). An investigation of the relationship between university rankings and
graduate starting wages. Journal of Institutional Research, 19(1), 46–54.
Deem, R., Mok, K. H., & Lucas, L. (2008). Transforming higher education in whose
image? Exploring the concept of the 'world-class' university in Europe and Asia.
Higher Education Policy 21(1), 83–97. doi: 10.1057/palgrave.hep.8300179
Delgado, J. E., & Weidman, J. C. (2012). Latin American and Caribbean countries in
the global quest for world class academic recognition: An analysis of publications in
Scopus and the Science Citation Index between 1990 and 2010. Excellence in Higher
Education 3, 111–121. doi: 10.5195/ehe.2012.73
Dill, D. D., & Soo, M. (2005). Academic quality, league tables, and public policy: A
cross-national analysis of university ranking systems. Higher Education, 49(4), 495–
533. doi: 10.1007/s10734-004-1746-8
Douglas, J. A., Ed. (2016). The new flagship university: Changing the paradigm from
global ranking to national relevancy. New York: Palgrave Macmillan.
Fidler, B., & Parsons, C. (2008). World university ranking methodologies: Stability and
variability. Higher Education Review, 40(3), 15–34.
Frey, B. S., & Rost, K. (2010). Do rankings reflect research quality? Journal of Applied
Economics, 13(1), 1–38.
Hazelkorn, E. (2009). Rankings and the battle for world-class excellence: Institutional
strategies and policy choices. Higher Education Management and Policy, 21(1), 55–
76.
Hazelkorn, E. (2014). Reflections on a decade of global rankings: What we've learned
and outstanding issues. European Journal of Education, 49(1), 12–28. doi:
10.1111/ejed.12059
Huang, M.-H. (2011). A comparison of three major academic rankings for world
universities: From a research evaluation perspective. Journal of Library and
Information Studies, 9(1), 1–25.
Huisman, J., & Currie, J. (2004). Accountability in higher education: Bridge over
troubled water? Higher Education, 48(4), 529–551.
Hou & Jacob
44
Kehm, B. M. (2014). Global university rankings—Impacts and unintended side effects.
European Journal of Education, 49(1), 102–112. doi: 10.1111/ejed.12064
Lee, M., & Park, H. W. (2012). Exploring the web visibility of world-class universities.
Scientometrics, 90(1), 201–218. doi: 10.1007/s11192-011-0515-6
Liu, N. C., & Cheng, Y. (2011). Global university rankings and their impact. In P. G.
Altbach (Ed.), Leadership for world-class universities: Challenges for developing
countries (pp. 145–158). New York and London: Routledge.
Lo, W. Y. W. (2011). Soft power, university rankings and knowledge production:
Distinctions between hegemony and self‐determination in higher education.
Comparative Education, 47(2), 209–222. doi: 10.1080/03050068.2011.554092
Marginson, S. (2014). University rankings and social science. European Journal of
Education, 49(1), 45–59. doi: 10.1111/ejed.12061
Marginson, S., & van der Wende, M. (2007). To rank or to be ranked: The impact of
global rankings in higher education. Journal of Studies in International Education,
11(3/4), 306–329. doi: 10.1177/1028315307303544
Proulx, R. (2007). Higher education ranking and leagues tables: Lessons learned from
benchmarking. Higher Education in Europe, 32(1), 71–82. doi:
10.1080/03797720701618898
QS Intelligence Unit (2015). How do students use rankings? The role of university
rankings in international student choice. London, UK: QS Intelligence Unit.
Quacquarelli Symonds. (2014). QS world university rankings 2013/14: Methodology.
Retrieved on January 19th, 2014 from http://www.iu.qs.com/university-
rankings/world-university-
rankings/?__hstc=238059679.bbf5e5a93ba9268580a1ed3ca7d6e2d7.1389930394437
.1389930394437.1389930394437.1&__hssc=238059679.2.1389930394438&__hsfp=
3348744073
Rauhvargers, A. (2011). Global university rankings and their impact. Belgium:
European University Association.
Roberts, D., & Thomson, L. (2007). Reputation management for universities: University
league tables and the impact on student recruitment. Leeds, UK: The Knowledge
Partnership.
Saisana, M., d’Hombres, B., & Saltelli, A. (2011). Rickety numbers: Volatility of
university rankings and policy implications. Research Policy, 40(1), 165–177. doi:
10.1016/j.respol.2010.09.003
Salmi, J., & Saroyan, A. (2007). League tables as policy instruments: Uses and misuses.
Higher education management and policy, 19(2), 31–68.
What contributes more to the ranking of higher education institutions?
45
Shanghai Jiao Tong University. (2014). Academic ranking of world universities 2013:
Methodology. Retrieved on January 19th, 2014 from
http://www.shanghairanking.com/ARWU-Methodology-2013.html
Shin, J. C., & Toutkoushian, R. K. (2011). The past, present, and future of university
rankings. In J. C. Shin, R. K. Toutkoushian & U. Teichler (Eds.), University
rankings: Theoretical basis, methodology and impacts on global higher education
(pp. 1–16). New York: Springer.
Su, S.-W. (2014). To be or not to be: Impact of “I” idolization from the perspectives of
humanities and social sciences faculty in Taiwan. In C. P. Chou (Ed.), The SSCI
syndrome in higher education: A local or global phenomenon (pp. 51–80).
Rotterdam, Netherlands: Sense Publishers.
Taylor, P., & Braddock, R. (2007). International university ranking systems and the idea
of university excellence. Journal of Higher Education Policy and Management,
29(3), 245–260. doi: 10.1080/13600800701457855
Thakur, M. (2007). The impact of ranking systems on higher rducation and its
stakeholders. Journal of Institutional Research, 13(1), 83–96.
Thomson Reuters. (2010). World university rankings 2010-2011: Methodology.
Retrieved on January 19th, 2014 from http://www.timeshighereducation.co.uk/world-
university-rankings/2010-11/world-ranking/methodology
Thomson Reuters. (2014). World university rankings 2013-2014: Methodology.
Retrieved on January 19th, 2014 from http://www.timeshighereducation.co.uk/world-
university-rankings/2013-14/world-ranking/methodology
Tofallis, C. (2012). A different approach to university rankings. Higher Education,
63(1), 1–18. doi: 10.1007/s10734-011-9417-z
Treiman, D. J. (2009). Quantitative data analysis: Doing social research to test ideas.
San Francisco, CA: Jossey-Bass.
Trigwell, K. (2011). Measuring teaching performance. In J. C. Shin, R. K. Toutkoushian
& U. Teichler (Eds.), University rankings: Theoretical basis, methodology and
impacts on global higher education (pp. 165–181). Dordrecht Heidelberg London
New York: Springer
van Raan, A. F. J. (2005). Fatal attraction: Conceptual and methodological problems in
the ranking of universities by bibliometric methods. Scientometrics, 62(1), 133–143.
doi: 10.1007/s11192-005-0008-6
van Vught, F. A., & Westerheijden, D. F. (2010). Multidimensional ranking: A new
transparency tool for higher education and research. Higher education management
and policy, 22(3), 31–56. doi: 10.1787/hemp-22-5km32wkjhf24
Vieira, R. C. & Lima, M. C. (2015). Academic ranking—from its genesis to its
international expansion. Higher Education Studies, 5(1), 63–72. doi:
10.5539/hes.v5n1p63
Hou & Jacob
46
Wildavsky, B. (2010). The great brain race: How global universities are reshaping the
world. Princeton, NJ: Princeton University Press.
Williams, R. (2008). Methodology, meaning and usefulness of rankings. Australian
Universities' Review, 50(2), 51–58.
Williams, R., & Van Dyke, N. (2008). Reputation and reality: Ranking major disciplines
in Australian universities. Higher Education, 56(1), 1–28. doi: 10.1007/s10734-007-
9086-0